How to Build a Community for Your Startup

Somebody told you to build a community. Probably a VC, probably on a podcast, probably right after they said "distribution is the moat." So you spun up a Slack workspace, invited 40 people from your waitlist, posted a welcome message, and then watched it turn into a room where you talk to yourself once a week.

That's the normal outcome. A Slack with 400 members and three messages a week isn't a growth channel. It's a ghost town with a logo on it.

But community building for startups does work, and when it works it compounds harder than any paid channel you'll ever run. The difference between the two outcomes isn't budget or headcount. It's whether the thing has a reason to exist that has nothing to do with your signup button.

Here's how to think about it.

What does "community" actually mean for a startup?

A startup community is a group of people who get value from each other, not just from you. That's the whole test. If every useful thing in the room comes from your team, you don't have a community. You have a support queue with extra steps.

This distinction matters more than it sounds. Most founders use "community" to mean one of four different things, and they need completely different work:

• An audience. People who consume what you publish. A newsletter, a YouTube channel, a Twitter following. Valuable, but one-to-many.
• A user group. Customers who talk to you about your product. Feature requests, bug reports, roadmap input. Useful, but it's really research.
• A support forum. Users answering each other's product questions. This one has clear ROI and it's the easiest to justify internally.
• An actual community. People who show up for each other, and your product happens to be the reason they met.

Only the last one is community-led growth. The other three are good things to build, and you should probably build them first, but don't confuse them. I've watched founders spend six months trying to force a support forum into being a movement, and it never works, because nobody wakes up wanting to hang out in a place designed to reduce your ticket volume.

Should you even build a community right now?

Probably not yet, and here's the test: can you name 20 people who would be disappointed if your community disappeared tomorrow? If not, you're too early.

Community is a retention and expansion channel dressed up as an acquisition channel. It works when you already have people who care. It does almost nothing when you're pre-product-market-fit and trying to manufacture demand out of an empty room.

Three signals that you're ready:

1. You have users who talk to each other already, in your DMs, in a group chat, in a Reddit thread you didn't start.
2. Your users have a shared problem that isn't your product. Founders share "how do I get my first customers." Notion users share "how do I build a CRM in this thing." That shared problem is the actual reason people show up.
3. Somebody on your team wants to do this. Not "has capacity." Wants to. Community work is unglamorous and relentless and it dies the second it becomes a chore assigned to whoever had a light week.

If you're missing all three, put the effort into customer acquisition channels that pay off faster and come back to this at a few hundred users.

Where do your first 100 members come from?

Your first members come from your existing users, hand-invited one at a time, over DM. Not a public link. Not a launch post. Individual invitations to specific people you already know are interested.

This feels absurdly slow. Do it anyway. The first 30 people set the culture permanently, and a community that opens to the public on day one gets its culture set by whoever wanders in.

Notion's ambassador program is the case study worth stealing from. They tweeted a call for people who wanted to get involved and got somewhere north of 600 applications in the first few weeks. They onboarded ten. Ten. That group went on to seed local chapters that now include a Vietnamese community with 200,000-plus members and a subreddit past 300,000. One Tokyo meetup pulled 300 attendees, and roughly 80% of them became paying users.

The lesson isn't "get 600 applications." It's that they treated the first cohort as a hiring decision rather than a signup form.

Practically, for your first 100:

• Go through your last three months of support conversations and pick the 20 people who were most enthusiastic or most specific.
• DM each one. Tell them what you're building and why you want them specifically. Name the thing they said or did that put them on the list.
• Ask each of those 20 to bring one person. That gets you to 40 with zero strangers.
• Run something in week one. A call, an AMA, a shared doc, a live teardown. Something with a time and a date, because an empty channel with no event is just a room.

Expect a 30% to 50% acceptance rate on personal DMs to warm users. If you're getting under 20%, the problem is the pitch, not the channel.

Slack, Discord, or something else?

Use Discord if your users are technical or under 35, use Slack if they're B2B professionals who already live in Slack all day, and don't agonize over it for more than an afternoon.

The practical differences that actually matter:

	Slack (free)	Discord (free)
Message history	Last 90 days visible, older content hidden	Unlimited
Cost to upgrade	Per member, around $7 to $9 per user monthly	Flat, Nitro is a per-user perk not a community fee
Search	Limited to the visible window on free	Full history
Culture	Feels like work	Feels like a hangout

That 90-day history limit is the one people underestimate. Slack's free plan hides messages past 90 days, which means every good answer your members write evaporates from search. For a community whose whole value is accumulated knowledge, that's a slow structural leak. Slack's paid plans price per member, so a community that grows to 2,000 people becomes a five-figure annual line item. Discord's economics don't break the same way.

There's a third option most founders skip: don't build a space at all. Run a recurring event instead. A monthly call, a cohort, a weekly office hours slot. Events create the same connections without requiring you to keep a room alive between them, and they're far easier to kill without embarrassment if the thing isn't working.

Why do most startup communities die within a year?

They die because the community was built as a marketing channel, so the moment it stops producing signups, it stops getting attention. Then engagement drops, then the founder posts less, then it's a ghost town, then nobody wants to be the one to admit it.

The specific failure patterns, in rough order of how often I see them:

Launching too wide. Public invite link, 500 members in a week, zero relationships. Big numbers, dead room. Portland Startups Slack is the well-documented version of this: over 6,500 members, and the people running it acknowledged most were lurkers or fully inactive.

Measuring member count. Member count is the community equivalent of registered users. It only goes up, it never tells you anything, and optimizing for it actively pushes you toward the wide-launch failure above.

No reason to return. If nothing happens on a schedule, people check once and forget. Recurring events are the heartbeat. Weekly, biweekly, monthly, pick one and never miss it.

Founder as the only node. If you're the answer to every question, the community collapses the week you get busy. Your job is to introduce people to each other and then get out of the way.

Treating it as free. It isn't. A functioning community needs somebody spending real hours on it every week. If nobody owns it, it will die, and it will take longer to admit than it should.

Here's the uncomfortable part. Commsor raised $16 million and then a $50 million Series B selling software for community-led growth, and later repositioned away from community entirely toward warm intros and networks. The company that most loudly sold the category concluded the category was harder to monetize than the pitch suggested. That's not a reason to skip community. It is a reason to be skeptical of anyone selling you a platform before you have ten people who care.

How do you measure whether a community is working?

Measure the percentage of members who did something in the last 30 days, and the percentage of conversations where the answer came from a member rather than your team. Those two numbers tell you almost everything.

The old rule of thumb is 90-9-1: 90% lurk, 9% react or reply occasionally, 1% create. It's a useful mental model and it's also contested. Higher Logic and others have published data arguing the split is outdated and that healthy communities beat it comfortably. Treat 90-9-1 as a floor rather than a law. If you're below it, something is broken.

A more useful scorecard for an early community:

• 30-day active rate. Members who posted, replied, or reacted in the last month, divided by total members. Under 10% and you have a list, not a community.
• Member-answered ratio. Of questions asked this month, what share got a good answer from someone who doesn't work for you? This is the single best leading indicator that the thing will survive without you.
• Time to first reply. How long does a new member's first post sit unanswered? Anything over a few hours in a small community is a culture problem, and it's fixable with three people whose explicit job is to reply fast.
• Second-post rate. What percentage of members who post once post again within 30 days? This catches bad first experiences that member count hides.

Track these in a spreadsheet, a Notion doc, or a planning tool like Foundra that keeps your go-to-market metrics next to the rest of your plan. The tool matters much less than the fact that you're looking at the same four numbers every month.

One more thing on measurement: give it 12 months before you judge it. Community is the slowest channel you will ever run. Lenny Rachitsky's subscriber Slack took years to reach the point where it was running nearly 200 real-world meetups across 30 countries, and it's now past 30,000 members. That growth is what compounding looks like from the outside. From the inside, for the first year, it looks like nothing.

What does community look like at 10, 100, and 1,000 members?

The job changes completely at each stage, and most founders keep doing the 10-member job at 500 members and wonder why it stopped scaling.

At 10 members, you are the community. Every conversation is you and one other person. Your only job is to make introductions between members who should know each other, and to be the most useful person in the room. No structure, no channels, no rules. One space.

At 100 members, your job is to create the reasons to show up. A recurring event, a weekly prompt, a shared resource people contribute to. Start naming the regulars publicly. Give the three most active members something official, even if it's just a role and a thank you. This is also where you add structure, and where you stop being able to reply to everything.

At 1,000 members, your job is to build the layer of people who run it without you. Moderators, chapter leads, ambassadors, whatever you call them. Notion's ambassadors and campus leads now run roughly 30 in-person events a month, which is not a thing any internal team could staff. That only works because the program hands real ownership to people outside the company.

The trap between 100 and 1,000 is trying to stay personally reachable. You can't, and pretending otherwise means new members get a worse experience than the early ones did, which is exactly backwards from what you want.

Key takeaways

• A community exists when members get value from each other. If all the value flows from you, it's a support channel.
• Don't start until you can name 20 people who'd miss it. Pre-PMF community building is almost always wasted effort.
• Hand-invite your first 30 members over DM. They set the culture permanently, and you only get one shot at it.
• Discord for technical or consumer audiences, Slack for B2B professionals, and check Slack's 90-day free history limit before you commit.
• Member count is a vanity metric. Track 30-day active rate and member-answered ratio instead.
• Recurring events are the heartbeat. Something on a schedule beats a room that's always open and always empty.
• Budget 12 months before you judge whether it's working, and budget real weekly hours from a named owner.

FAQ

How many members does a startup community need to be worth it?

Around 100 engaged members is where it starts generating value on its own. Below that you're the only source of activity. But engagement matters far more than size: 100 people where 30 post monthly beats 2,000 people where 20 do.

How much time does running a startup community take?

Plan for 5 to 10 hours a week at the 100-member stage, and expect it to be founder time early on. It drops per member as you promote moderators, but it never goes to zero. If nobody has those hours, don't start.

Should my community be free or paid?

Free, unless the community itself is the product. Paid communities filter for commitment and cut noise, but they also cap your growth and add an expectation of service delivery. Most startups are better off making the community a benefit of using the product.

What's the difference between community-led growth and just having a Discord?

Community-led growth means the community produces measurable outcomes: referrals, retention, support deflection, product feedback. Having a Discord means you have a Discord. The difference is whether anyone owns it and whether anyone measures it.

How do I restart a community that's already dead?

Archive it and start over with a different format, usually a recurring event instead of a room. Trying to resuscitate a dead channel rarely works, because new members judge activity by what they see in the scrollback, and the scrollback is the problem.

Can I build a community before I have a product?

Yes, and it's one of the few cases where community makes sense early. If you're building an audience around a problem rather than a product, the community becomes your validation channel and your first customer list. Just be clear with yourself that you're building an audience, and that audiences convert far worse than you expect.